[python] Migrate to parametrize_with_checks for scikit-learn integration tests

[rth]

Minor maintenance point but it could be worth migrating to parametrize_with_checks instead of using check_estimator directly in scikit-learn integration tests (or rather the private _yield_all_checks function from scikit-learn):

LightGBM/tests/python_package_test/test_sklearn.py

Lines 285 to 292 in c870c57

	def test_sklearn_integration(self):
	# we cannot use `check_estimator` directly since there is no skip test mechanism
	for name, estimator in ((lgb.sklearn.LGBMClassifier.__name__, lgb.sklearn.LGBMClassifier),
	(lgb.sklearn.LGBMRegressor.__name__, lgb.sklearn.LGBMRegressor)):
	check_parameters_default_constructible(name, estimator)
	# we cannot leave default params (see https://github.com/microsoft/LightGBM/issues/833)
	estimator = estimator(min_child_samples=1, min_data_in_bin=1)
	for check in _yield_all_checks(name, estimator):

It particular, this would make each pair of estimator-check an independent pytest check which will make them easier to skip or debug issues like #2628

Reference : https://scikit-learn.org/stable/developers/develop.html#rolling-your-own-estimator

[StrikerRUS]

Thanks for letting us know about parametrize_with_checks! I believe that with estimator tags mechanism it'll simplify some code.

I'm adding this to our requests hub.

[StrikerRUS]

Closed in favor of being in #2302. We decided to keep all feature requests in one place.

Welcome to contribute this feature! Please re-open this issue (or post a comment if you are not a topic starter) if you are actively working on implementing this feature.